Poetry is useful for dependency management AND packaging.
MUST be installed through curl. Any other way may cause problems on the future.
Create a new python package: poetry new [name]. It generates a project skeleton.
You can use poetry for development, and pip for production.
poetry init -n only generates pyproject.toml.
poetry install installs the project dependencies into a new virtualenv and generates poetry.lock. By default it installs the development libraries. For production, you must install as you install your packages normally (e.g. with pipx or as a dependency for other project).
poetry show --tree : shows all dependencies.
install a new dependency: poetry add [name]
install a new dev dependency: poetry add --dev [name]
remove a dependency: poetry remove [name]
remove a dev dependency: poetry remove --dev [name]
How to define as a script to be run:
[tool.poetry.scripts]
bagulho-cli = 'bagulho.cli:cli' # package.script:main_function
generate the package and wheel: poetry build
publicar no PyPI: poetry publish
open a shell (to run pytest, your script, etc…): poetry shell
update dependencies: poetry update
update poetry: poetry self update
generate a requirements file (only the prod, not dev ddependencies): poetry export
If you need virtualenv cache (e.g. on CI pipelines), it will be a little hard to do that with poetry.
You can point another PyPi-like packages repository to poetry.